Skip to content

Home

Jump to bottom
savvy773 edited this page May 26, 2026 · 3 revisions

AI Usage Dashboard

Monitor Claude, Codex, and Gemini CLI usage in one local dashboard.
Runs entirely on your machine — no cloud, no telemetry, no API keys.

Quick Navigation

Page Description
Quick Start Install and run in under 2 minutes
Architecture How the server, collector, and parser work together
Fix Checklist Step-by-step diagnostics when a provider shows wrong data
API Reference REST endpoints and payload shapes

What It Does

The browser never runs CLIs directly. A SvelteKit server API spawns each CLI in a virtual terminal via node-pty, captures the output, parses usage data, and writes results to data/usage-history.json. The UI renders from cached JSON on first paint, then refreshes in the background.

browser  →  GET /api/usage          →  render cached JSON
         →  POST /api/usage/refresh →  start collection
         →  poll GET /api/usage     →  update when done

server   →  spawn CLI in node-pty
         →  send slash command after readiness
         →  capture + normalize terminal output
         →  parse provider usage
         →  write usage-history.json / usage-latest.json

Providers

Provider Slash Command What's Shown
Claude /usage Current session + weekly usage
Codex /status 5h limit + weekly limit
Gemini CLI /model Per-model usage rates + resets

Links

Clone this wiki locally